Oxygen saturation regularity analysis in the diagnosis of obstructive sleep apnea

del Campo, Félix; Hornero, Roberto; Zamarrón, Carlos; Abasolo, Daniel E.; Álvarez, Daniel

doi:10.1016/j.artmed.2005.10.005

« Previous Next »Artificial Intelligence in Medicine
Volume 37, Issue 2 , Pages 111-118, June 2006

Oxygen saturation regularity analysis in the diagnosis of obstructive sleep apnea

Félix del Campo
- Hospital del Río Hortega, Servicio de Neumología, C/Cardenal Torquemada s/n, 47010 Valladolid, Spain
- Corresponding author. Present address: Padre Francisco Suárez 8, 5° dcha, 47006 Valladolid, Spain. Tel.: +34 606 983 938; fax: +34 983 331 566.
Roberto Hornero
- ETSI-Telecomunicación University of Valladolid, Camino del Cementerio s/n, 47011 Valladolid, Spain
Carlos Zamarrón
- Hospital Clínico Universitario, Servicio de Neumología, C/Travesia de La Choupana s/n, 15706 Santiago de Compostela, Spain
Daniel E. Abasolo
- ETSI-Telecomunicación University of Valladolid, Camino del Cementerio s/n, 47011 Valladolid, Spain
Daniel Álvarez
- ETSI-Telecomunicación University of Valladolid, Camino del Cementerio s/n, 47011 Valladolid, Spain

Received 1 June 2005; received in revised form 10 October 2005; accepted 15 October 2005.

Summary

Objective

The present study assessed the validity of approximate entropy (ApEn) analysis of arterial oxygen saturation (SaO₂) data obtained from pulse oximetric recordings as a diagnostic test for obstructive sleep apnea (OSA) in patients clinically suspected of suffering this disease.

Methodology

A sample of 187 referred outpatients, clinically suspected of having OSA, was studied using nocturnal pulse oximetric recording performed simultaneously with complete polysomnography. ApEn analysis was applied to SaO₂ data.

Results

Patients with OSA presented significantly higher approximate entropy levels than those without OSA (1.08±0.30 versus 0.47±0.26). Apnea–hypopnea index was correlated significantly with ApEn (r=0.607; p<0.001). Using receiver operating characteristic curve analysis, we obtained a diagnostic sensitivity of 88.3% and specificity of 82.9%, positive predictive value of 88.3% and a negative predictive value of 82.9%, at a threshold of 0.679. As a diagnostic test, this method presents high sensitivity and specificity compared to traditional methods in the diagnosis of OSA.